From fcd01f885163de8d9bcedfdb4db88b2d523b819c Mon Sep 17 00:00:00 2001 From: BlinkDL Date: Thu, 28 Oct 2021 15:16:45 +0800 Subject: [PATCH] no message --- train.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/train.py b/train.py index 61947dd..1e0b2bd 100644 --- a/train.py +++ b/train.py @@ -41,6 +41,8 @@ epoch_save_path = 'trained-' batch_size = 48 # if you see "CUDA out of memory", reduce this. # if you have good GPU, increase this. # use GPU-Z to find the highest value for your VRAM. + +n_epoch = 100 # the 'epoch' here is actually very short (and of fixed length) ######################################################################################## model_level = 'character' # 'character' (recommended) or 'word' @@ -52,7 +54,6 @@ n_embd = n_head * 64 n_attn = n_embd n_ffn = n_embd -n_epoch = 50 # the 'epoch' here is actually very short (and of fixed length) lr_init = 8e-4 if model_type == 'RWKV' else 4e-4 # RWKV can use higher lr lr_final = 1e-5